Assessing crisis management effectiveness: an application of a network autocorrelation model for multi-level networks

Angela Guerrero, Orjan Bodin, Daniel Nohrstedt, Lorien Jasny


During major disasters and emergencies, such as an escalating wildfire, actors from different sectors of society (government, volunteer organisations, emergency services etc) come together to collectively respond forming a “crisis response network”. In such networks, each actor (an individual or organisation) is faced with one or multiple complex tasks that not one actor can carry out by themselves (e.g. evacuation, fire extinction, logistics and supply, public communication etc.)—thus they need to collaborate with others. The tasks can also be interdependent (e.g. fire extinction cannot be done without appropriate logistics). Thus, the performance of actors not only depends on how well they are doing themselves, but also on their collaborators’ performance on the same task, and on how well interdependent tasks are addressed by others in the responder network. We analyse data from crisis responders of large-scale wildfires in Sweden and Canada using a network autocorrelation model to test the impact of specific patterns of collaboration and task interdependency on actors’ performance, while accounting for other actor level effects such as experience, organisational affiliation and level of professionalization. Our model is different from existing network autocorrelation models in two main ways. First, we are modelling using multilevel graphs (as opposed to one-mode graphs), and thus are able to make use of the rich information provided by our multilevel actor-task data structure. Second, since our theoretical attention is on actors’ performance in addressing specific tasks, we use the model to predict tie weights (as opposed to actor/node-level attributes). Early results indicate that different patterns of actor-task entanglements significantly impact performance

← Schedule